[ 
https://issues.apache.org/jira/browse/LUCENE-5329?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13814197#comment-13814197
 ] 

Areek Zillur commented on LUCENE-5329:
--------------------------------------

These are my thoughts on this:
  - From the Lucene Suggester perspective, it makes perfect sense to have the 
payload as a binary field. Because the payload is just 'stored' as is and then 
returned, no processing is done on it, hence it makes sense that the lucene 
suggester just treats it as binary data.
  - Having said that when exposing it from Solr, it would be nice to make it 
"just work", rather than the user having to make sure what field will spit what 
out, hence the proposed changes
  - Regarding the use-case with ecommerce store, the payload does not 
necessarily have  to be a field, it can be the aggregate of other fields or 
some arbitrary associated data. (though there is no way to do so in Solr now, 
but I plan to make it possible with the new Solr Suggester (SOLR-5378) :).
  - As far as I understand, payloads should remain binary in the Lucene 
Suggesters, but the dilemma is whether the input to the suggester be flexible 

> Make DocumentDictionary and co more lenient to dirty documents
> --------------------------------------------------------------
>
>                 Key: LUCENE-5329
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5329
>             Project: Lucene - Core
>          Issue Type: Improvement
>          Components: core/search
>            Reporter: Areek Zillur
>         Attachments: LUCENE-5329.patch
>
>
> Currently DocumentDictionary errors out whenever any document does not have 
> value for any relevant stored fields. It would be nice to make it lenient and 
> instead ignore the invalid documents.
> Another "issue" with the DocumentDictionary is that it only allows string 
> fields as suggestions and binary fields as payloads. When exposing these 
> dictionaries to solr (via https://issues.apache.org/jira/browse/SOLR-5378), 
> it is inconvenient for the user to ensure that a suggestion field is a string 
> field and a payload field is a binary field. It would be nice to have the 
> dictionary "just work" whenever a string/binary field is passed to 
> suggestion/payload field. The patch provides one solution to this problem (by 
> accepting string or binary values), though it would be great if there are any 
> other solution to this, without making the DocumentDictionary "too flexible"



--
This message was sent by Atlassian JIRA
(v6.1#6144)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to